Pancreatic cancer is relentless: Nearly all of the 30,000 Americans diagnosed annually with the disease die within 12 months. The early symptoms, back pain and indigestion, are so vague that most patients have no idea that they have cancer. By the time it's detected, the disease has usually spread to the point that it is untreatable.
But what if a simple blood test could alert doctors to pancreatic cancer early enough to treat it? Such a test does not exist, but University of Nebraska researcher Michael Hollingsworth thinks he has a solid lead. He has found that the disease increases levels of proteins called mucins.
"That may be a warning signal that you have a developing case of pancreatic cancer," says Hollings- worth, who is measuring mucins in the lab to test his theory.
Hollingsworth is part of a new trend in the fight against cancer: the search for "biomarkers," specific molecules - usually proteins or genetic material - that offer early warning signs of disease. Rather than looking for ways to attack cancer once it's established, biomarker researchers want to catch it when it is more susceptible to treatment.
"We know that cancer takes years to form," says Johns Hopkins University cancer specialist Dr. David Sidransky, a leading biomarker researcher. "We want to treat the tumors before they're cancer."
Sidransky and his team are trying to find biomarkers for cancers of the bladder, kidney, prostate, lung and esophagus.
The new approach is increasingly popular: All the major pharmaceutical makers and many smaller biotech companies are looking for cancer biomarkers, and the National Cancer Institute has started a $32 million program to encourage biomarker research.
"It's a hot area," says Dr. Pierre Massion, a biomarker researcher at Vanderbilt University in Nashville, Tenn. "People realize that trying to treat advanced cancer is very hard."
Massion is trying to find biomarkers for lung cancer, which kills about 160,000 people a year in the United States. In 80 percent of those patients, the disease is discovered too late to surgically remove the tumors - the cancer has spread to lymph nodes and other organs.
Like many researchers, Massion is looking for clues among the tens of thousands of proteins that travel in the bloodstream. When cancer appears, it can raise or lower levels of some of these molecules, offering a potential early clue.
With the advent of technology that enables scientists to analyze thousands of proteins at a time, finding these telltale molecules has become easier in recent years. Using a technique called mass spectrometry, which sorts proteins by weight, researchers can determine with relative ease how various cancers affect the levels of hundreds or thousands of proteins.
Other technologies are also proving useful: For example, techniques developed for the Human Genome Project enable scientists to examine thousands of genes for possible links to cancer.
The biomarker concept is not new. Doctors have used the approach for decades, but most current biomarkers have serious flaws. Some do not detect cancer until too late, while others are not accurate enough, showing positive in patients without cancer, or worse, negative in those who are sick.
The best-known biomarker is prostate-specific antigen, or PSA, widely used to test for prostate cancer. Most doctors say the test can be helpful, but it is often confusing and inexact.
Though a high PSA level can indicate prostate cancer, only about 25 percent of men with high PSA levels have the disease. At the same time, a low PSA level does not guarantee health - 15 percent of men with low levels turn out to have the disease.
Hopkins pathologist Robert H. Getzenberg hopes to make the PSA test obsolete. For the past 10 years, he has studied a protein called early prostate cancer antigen (EPCA). In a small study published last year, he found that the protein predicted prostate cancer with 90 percent accuracy; that is, nine out of 10 men with increased EPCA levels in their blood went on to develop the disease.
Getzenberg has just finished a much larger study and says the results are very promising. He predicts that the test will be able to detect prostate cancer up to two or three years before tumors appear. If his work pans out, it could transform the way the disease is treated.
"We're going to be much better at early detection," he says. "We can focus on the men who actually have the disease."
Scientists elsewhere are conducting human trials for other biomarkers. At the University of Michigan, oncologist Dr. Dean Brenner is targeting colon and liver cancers. At Hopkins, Sidransky's group is involved in a nationwide study of 500 volunteers to examine a DNA-based biomarker for bladder cancer.
And at the University of Maryland, gastroenterologists Drs. Stephen J. Meltzer and Bruce D. Greenwald are following a group of esophageal cancer patients, as well as some who are at high risk of developing the disease. They check the patients for a special form of gene abnormality called methylated DNA.
The subjects receive regular blood tests and biopsies to measure levels of methylated DNA for several genes that play a role in suppressing tumors. In animal and human tests, rising levels of these methylated genes seem to predict cancer up to two years before tumors appear.
"That gives you a two-year window to intervene," Meltzer says.
That window is very important to retired Bel Air schoolteacher Bernette Henry, who is taking part in the study. Four years ago, she was diagnosed with esophageal cancer. Surgeons removed two-thirds of her esophagus, apparently catching the tumor before it had a chance to spread.
But Henry, 64, worries about a recurrence, particularly because so much of her esophagus is gone that more surgery would not be possible.
"That would be phenomenal if they can catch it early," she says.
Drug companies are equally excited about biomarkers: An accurate cancer test used by millions of patients could produce huge profits.
"Almost everybody is in this market in one way or another," says Sylvia Cerqueira, a health-care research analyst for Frost & Sullivan, a consulting firm in San Antonio. "It's a race to find these things."
Roche, for example, is conducting pilot human studies of a test to detect changes in P53, a gene that seems to suppress tumors. In many cancers, the gene somehow switches off, allowing tumors to proliferate. Tom Metcalfe, head of the drug company's new biomarker program, says the test might eventually help spot many kinds of cancers.
Many companies are seeking biomarkers that not only detect cancer but reveal details about the particular genetic strain of illness. With such tests, doctors can avoid guesswork, tailoring treatment to attack specific subtypes of the disease.
"We need biomarkers that tell us how the cancer will behave," says Sudhir Srivastava, director of the National Cancer Institute's biomarker program.
This "targeted" approach has been successful. Since 1998, many breast cancer patients have been tested for the HER2 gene, which may play a role in tumor growth. Women whose biopsies reveal high levels of HER2 are usually treated with Herceptin, a drug that keeps the gene from generating growth.
The approach seems to work: Two large studies published last fall found that HER2-positive patients treated with Herceptin tended to have fewer recurrences than women who did not receive the drug.
"We will have more drugs along the Herceptin model," says Roche's Metcalfe. "We will be able to more rationally predict which patients will respond to which drugs."
Eventually, biomarkers will move beyond detecting cancers to predicting them years in advance, Sidransky says. He envisions a day when a blood test will tell patients that they have a high risk of getting a certain cancer in coming decades. With so much advance warning, they might be able to avoid the illness with nothing more than a healthy diet and regular exercise.
"Your treatment options will be many," he says, "and your outcome will be great."